1
|
Fanizzi A, Arezzo F, Cormio G, Comes MC, Cazzato G, Boldrini L, Bove S, Bollino M, Kardhashi A, Silvestris E, Quarto P, Mongelli M, Naglieri E, Signorile R, Loizzi V, Massafra R. An explainable machine learning model to solid adnexal masses diagnosis based on clinical data and qualitative ultrasound indicators. Cancer Med 2024; 13:e7425. [PMID: 38923847 PMCID: PMC11196372 DOI: 10.1002/cam4.7425] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 10/09/2023] [Revised: 06/10/2024] [Accepted: 06/11/2024] [Indexed: 06/28/2024] Open
Abstract
BACKGROUND Accurate characterization of newly diagnosed a solid adnexal lesion is a key step in defining the most appropriate therapeutic approach. Despite guidance from the International Ovarian Tumor Analyzes Panel, the evaluation of these lesions can be challenging. Recent studies have demonstrated how machine learning techniques can be applied to clinical data to solve this diagnostic problem. However, ML models can often consider as black-boxes due to the difficulty of understanding the decision-making process used by the algorithm to obtain a specific result. AIMS For this purpose, we propose an Explainable Artificial Intelligence model trained on clinical characteristics and qualitative ultrasound indicators to predict solid adnexal masses diagnosis. MATERIALS & METHODS Since the diagnostic task was a three-class problem (benign tumor, invasive cancer, or ovarian metastasis), we proposed a waterfall classification model: a first model was trained and validated to discriminate benign versus malignant, a second model was trained to distinguish nonmetastatic versus metastatic malignant lesion which occurs when a patient is predicted to be malignant by the first model. Firstly, a stepwise feature selection procedure was implemented. The classification performances were validated on Leave One Out scheme. RESULTS The accuracy of the three-class model reaches an overall accuracy of 86.36%, and the precision per-class of the benign, nonmetastatic malignant, and metastatic malignant classes were 86.96%, 87.27%, and 77.78%, respectively. DISCUSSION SHapley Additive exPlanations were performed to visually show how the machine learning model made a specific decision. For each patient, the SHAP values expressed how each characteristic contributed to the classification result. Such information represents an added value for the clinical usability of a diagnostic system. CONCLUSIONS This is the first work that attempts to design an explainable machine-learning tool for the histological diagnosis of solid masses of the ovary.
Collapse
Affiliation(s)
- Annarita Fanizzi
- Laboratorio Biostatistica e BioinformaticaI.R.C.C.S. Istituto Tumori ‘Giovanni Paolo II’BariItaly
| | - Francesca Arezzo
- Gynecologic Oncology UnitIRCCS Istituto Tumori “Giovanni Paolo II”BariItaly
- Department of Precision and Regenerative Medicine – Ionian AreaUniversity of Bari “Aldo Moro”BariItaly
| | - Gennaro Cormio
- Gynecologic Oncology UnitIRCCS Istituto Tumori “Giovanni Paolo II”BariItaly
- Interdisciplinar Department of MedicineUniversity of Bari “Aldo Moro”BariItaly
| | - Maria Colomba Comes
- Laboratorio Biostatistica e BioinformaticaI.R.C.C.S. Istituto Tumori ‘Giovanni Paolo II’BariItaly
| | - Gerardo Cazzato
- Section of Molecular Pathology, Department of Emergency and Organ TransplantationUniversity of Bari “Aldo Moro”BariItaly
| | - Luca Boldrini
- Fondazione Policlinico Universitario “A. Gemelli” IRCCSItaly
| | - Samantha Bove
- Laboratorio Biostatistica e BioinformaticaI.R.C.C.S. Istituto Tumori ‘Giovanni Paolo II’BariItaly
| | - Michele Bollino
- Department of Obstetrics and Gynecology, Division of Gynecologic oncology, Skåne University Hospital and Lund UniversityFaculty of Medicine, Clinical SciencesLundSweden
| | - Anila Kardhashi
- Gynecologic Oncology UnitIRCCS Istituto Tumori “Giovanni Paolo II”BariItaly
| | - Erica Silvestris
- Gynecologic Oncology UnitIRCCS Istituto Tumori “Giovanni Paolo II”BariItaly
| | - Pietro Quarto
- Gynecologic Oncology UnitIRCCS Istituto Tumori “Giovanni Paolo II”BariItaly
- Interdisciplinar Department of MedicineUniversity of Bari “Aldo Moro”BariItaly
| | - Michele Mongelli
- Department of Precision and Regenerative Medicine – Ionian AreaUniversity of Bari “Aldo Moro”BariItaly
| | - Emanuele Naglieri
- Medical Oncology Unit, IRCCSIstituto Tumori Giovanni Paolo IIBariItaly
| | - Rahel Signorile
- Laboratorio Biostatistica e BioinformaticaI.R.C.C.S. Istituto Tumori ‘Giovanni Paolo II’BariItaly
| | - Vera Loizzi
- Gynecologic Oncology UnitIRCCS Istituto Tumori “Giovanni Paolo II”BariItaly
- Interdisciplinar Department of MedicineUniversity of Bari “Aldo Moro”BariItaly
| | - Raffaella Massafra
- Laboratorio Biostatistica e BioinformaticaI.R.C.C.S. Istituto Tumori ‘Giovanni Paolo II’BariItaly
| |
Collapse
|
2
|
Asp M, Mockute I, Måsbäck A, Liuba K, Kannisto P, Malander S. Tru-Cut Biopsy in Gynecological Cancer: Adequacy, Accuracy, Safety and Clinical Applicability. J Multidiscip Healthc 2023; 16:1367-1377. [PMID: 37215751 PMCID: PMC10198176 DOI: 10.2147/jmdh.s396788] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/16/2022] [Accepted: 04/18/2023] [Indexed: 05/24/2023] Open
Abstract
Purpose Tru-cut biopsy is a minimally invasive technique used to obtain tissue samples for the diagnosis of tumors, especially in patients where primary surgery is not indicated. The aim of this study was to assess the adequacy, accuracy and safety of the tru-cut biopsy for diagnosis in gynecological cancer. Methods A retrospective population-based review of 328 biopsies was conducted. The indications for tru-cut biopsies were diagnosis of primary tumors, metastases of gynecological and non-gynecological tumors, and suspected recurrences. A tissue sample was considered adequate when the quality/quality was sufficient to identify the subtype/origin of the tumor. Potential factors affecting adequacy were analyzed using logistic regressions analyses. Accuracy was defined as agreement between the diagnosis of the tru-cut biopsy and the postoperative histology. The therapy plan was registered, and the clinical applicability of the tru-cut biopsy was investigated. Complications within 30 days after the biopsy procedure were registered. Results In total, 300 biopsies were identified as tru-cut biopsies. The overall adequacy was 86.3%, varying between 80.8% and 93.5%, respectively, when performed by a gynecological oncologist or a gynecologist with a subspecialty in ultrasound diagnosis. Sampling of a pelvic mass had a lower adequacy (81.6%) compared with sampling of the omentum (93.9%) or carcinomatosis (91.5%). The accuracy was 97.5%, and the complication rate was 1.3%. Conclusion The tru-cut biopsy is a safe and reliable diagnostic method with a high accuracy and a good adequacy, depending on the site of the tissue sample, indications for the biopsy and the experience of the operator.
Collapse
Affiliation(s)
- Mihaela Asp
- Department of Obstetrics and Gynecology, Department of Clinical Science, Skåne University Hospital, Lund University, Lund, Sweden
| | - Ingrida Mockute
- Department of Obstetrics and Gynecology, Department of Clinical Science, Skåne University Hospital, Lund University, Lund, Sweden
| | - Anna Måsbäck
- Department of Clinical Genetics and Pathology, Department of Clinical Science, Skåne University Hospital, Lund University, Lund, Sweden
| | - Karina Liuba
- Department of Obstetrics and Gynecology, Department of Clinical Science, Skåne University Hospital, Lund University, Lund, Sweden
| | - Päivi Kannisto
- Department of Obstetrics and Gynecology, Department of Clinical Science, Skåne University Hospital, Lund University, Lund, Sweden
| | - Susanne Malander
- Department of Oncology and Pathology, Department of Clinical Science, Skåne University Hospital, Lund University, Lund, Sweden
| |
Collapse
|
3
|
Pelayo-Delgado I, Sancho J, Pelayo M, Corraliza V, Perez-Mies B, Del Valle C, Abarca L, Pablos MJ, Martin-Gromaz C, Pérez-Vidal JR, Penades I, Garcia E, Llanos MC, Alcazar JL. Contribution of Outpatient Ultrasound Transvaginal Biopsy and Puncture in the Diagnosis and Treatment of Pelvic Lesions: A Bicenter Study. Diagnostics (Basel) 2023; 13:diagnostics13030380. [PMID: 36766484 PMCID: PMC9913928 DOI: 10.3390/diagnostics13030380] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/19/2022] [Revised: 01/09/2023] [Accepted: 01/17/2023] [Indexed: 01/20/2023] Open
Abstract
BACKGROUND The use of transvaginal ultrasound guided biopsy and puncture of pelvic lesions is a minimally invasive technique that allows for accurate diagnosis. It has many advantages compared to other more invasive (lower complication rate) or non-invasive techniques (accurate diagnosis). Furthermore, it offers greater availability, it does not radiate, enables the study of pelvic masses accessible vaginally with ultrasound control in real time, and it is possible to use the colour Doppler avoiding puncturing large vessels among others. The main aim of the work is to describe a standardized ambulatory technique and to determine its usefulness. METHODS This is a retrospective study of ultrasound transvaginal punctures (core needle biopsies and cytologies) and drainages of pelvic lesions performed on an outpatient basis during the last two years. The punctures were made with local anesthesia, under transvaginal ultrasound guidance with an automatic or semi-automatic 18G biopsy needle with a length of 20-25 cm and a penetration depth of 12 or 22 mm. The material obtained was sent for anatomopathological, cytological and/or microbiological study if necessary. RESULTS A total of 42 women were recruited in two centers. Fifty procedures (nine punctures, seven drains, and 34 biopsies) were performed. In five cases the punction and drain provided clinical relief in benign pelvic masses. Regarding material of the biopsies performed, 15 were vaginal in women previously histerectomized, finding 10 carcinomas, eight were ovarian tumours in advanced stages or peritoneal carcinomatosis obtaining the appropriate histology in each case, seven were suspicious cervical biopsies finding carcinomas in five of them, three were myometrial biopsies including one breast carcinoma metastasis in the miometrium and a benign placental nodule, and a periurethral biopsy was performed on a woman with a history of endometrial cancer confirming recurrence. The pathological diagnosis was satisfactory in all cases, confirming the nature of the lesion (25 malignant-ten vaginal recurrences of previous gynaecological cancers, eight cases of primary ovarian/peritoneal carcinoma, four new diagnosis of cervical malignant masses, one cervical metastasis of lymphoma, one periurethral recurrence of endometrial carcinoma and one recurrence of breast cancer in the myometrium-and 23 benign). The tolerance was excellent and no complications were detected. CONCLUSION The ambulatory ultrasound transvaginal puncture and drainage technique is useful for obtaining a sample for pathological and microbiological diagnosis with excellent tolerance that can be used to rule out the recurrence of malignant lesions or progression of the disease, diagnose masses not accessible to gynecological exploration (vaginal vault, myometrium or cervix) and for early histologic diagnosis in cases of advanced peritoneal carcinomatosis or ovarian carcinoma as well as drainage and cytological study of cystic pelvic masses.
Collapse
Affiliation(s)
- Irene Pelayo-Delgado
- Department of Obstetrics and Gynecology, University Hospital Ramón y Cajal, Alcalá de Henares University, 28034 Madrid, Spain
| | - Javier Sancho
- Department of Obstetrics and Gynecology, University Hospital Ramón y Cajal, Alcalá de Henares University, 28034 Madrid, Spain
| | - Mar Pelayo
- Department of Radiology, Hospital HM Puerta del Sur. Hospital HM Rivas, 28938 Madrid, Spain
| | - Virginia Corraliza
- Department of Obstetrics and Gynecology, University Hospital Ramón y Cajal, Alcalá de Henares University, 28034 Madrid, Spain
| | - Belen Perez-Mies
- Department of Pathology, University Hospital Ramón y Cajal, 28034 Madrid, Spain
| | - Cristina Del Valle
- Department of Obstetrics and Gynecology, University Hospital Ramón y Cajal, Alcalá de Henares University, 28034 Madrid, Spain
| | - Leopoldo Abarca
- Department of Obstetrics and Gynecology, University Hospital Ramón y Cajal, Alcalá de Henares University, 28034 Madrid, Spain
| | - Maria Jesus Pablos
- Department of Obstetrics and Gynecology, University Hospital Ramón y Cajal, Alcalá de Henares University, 28034 Madrid, Spain
| | - Carmen Martin-Gromaz
- Department of Obstetrics and Gynecology, University Hospital Ramón y Cajal, Alcalá de Henares University, 28034 Madrid, Spain
| | - Juan Ramón Pérez-Vidal
- Department of Obstetrics and Gynecology, University Hospital Virgen de la Arrixaca, 30120 Murcia, Spain
| | - Inmaculada Penades
- Department of Obstetrics and Gynecology, University Hospital Virgen de la Arrixaca, 30120 Murcia, Spain
| | - Elvira Garcia
- Department of Obstetrics and Gynecology, University Hospital Virgen de la Arrixaca, 30120 Murcia, Spain
| | - Maria Carmen Llanos
- Department of Obstetrics and Gynecology, University Hospital Virgen de la Arrixaca, 30120 Murcia, Spain
| | - Juan Luis Alcazar
- Department of Obstetrics and Gynecology, Clinica Universidad de Navarra, 31008 Pamplona, Spain
- Correspondence:
| |
Collapse
|
4
|
Fischerova D, Scovazzi U, Sousa N, Hovhannisyan T, Burgetova A, Dundr P, Němejcová K, Bennett R, Vočka M, Frühauf F, Kocian R, Indrielle-Kelly T, Cibula D. Primary retroperitoneal nodal endometrioid carcinoma associated with Lynch syndrome: A case report. Front Oncol 2023; 13:1092044. [PMID: 36895475 PMCID: PMC9989303 DOI: 10.3389/fonc.2023.1092044] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/07/2022] [Accepted: 01/20/2023] [Indexed: 02/25/2023] Open
Abstract
We report a rare case of primary nodal, poorly differentiated endometrioid carcinoma associated with Lynch syndrome. A 29-year-old female patient was referred by her general gynecologist for further imaging with suspected right-sided ovarian endometrioid cyst. Ultrasound examination by an expert gynecological sonographer at tertiary center revealed unremarkable findings in the abdomen and pelvis apart from three iliac lymph nodes showing signs of malignant infiltration in the right obturator fossa and two lesions in the 4b segment of the liver. During the same appointment ultrasound guided tru-cut biopsy was performed to differentiate hematological malignancy from carcinomatous lymph node infiltration. Based on the histological findings of endometrioid carcinoma from lymph node biopsy, primary debulking surgery including hysterectomy and salpingo-oophorectomy was performed. Endometrioid carcinoma was confirmed only in the three lymph nodes suspected on the expert scan and primary nodal origin of endometroid carcinoma developed from ectopic Müllerian tissue was considered. As a part of the pathological examination immunohistochemistry analysis for mismatch repair protein (MMR) expression was done. The findings of deficient mismatch repair proteins (dMMR) led to additional genetic testing, which revealed deletion of the entire EPCAM gene up to exon 1-8 of the MSH2 gene. This was unexpected considering her insignificant family history of cancer. We discuss the diagnostic work-up for patients presenting with metastatic lymph node infiltration by cancer of unknown primary and possible reasons for malignant lymph node transformation associated with Lynch syndrome.
Collapse
Affiliation(s)
- Daniela Fischerova
- Department of Obstetrics and Gynecology, First Faculty of Medicine, Charles University and General University Hospital in Prague, Prague, Czechia
| | - Umberto Scovazzi
- Department of Gynecology and Obstetrics, Ospedale Policlinico San Martino and University of Genoa, Genova, Italy
| | - Natacha Sousa
- Department of Gynecology and Obstetrics, Hospital de Braga, Braga, Portugal
| | - Tatevik Hovhannisyan
- Department of Gynecology and Gynecologic Oncology, Nairi Medical Center (MC), Yerevan, Armenia
| | - Andrea Burgetova
- Department of Radiology, First Faculty of Medicine, Charles University and General University Hospital in Prague, Prague, Czechia
| | - Pavel Dundr
- Department of Pathology, First Faculty of Medicine, Charles University and General University Hospital in Prague, Prague, Czechia
| | - Kristýna Němejcová
- Department of Pathology, First Faculty of Medicine, Charles University and General University Hospital in Prague, Prague, Czechia
| | - Rosalie Bennett
- Department of Pathology, First Faculty of Medicine, Charles University and General University Hospital in Prague, Prague, Czechia
| | - Michal Vočka
- Department of Oncology, First Faculty of Medicine, Charles University, Prague, Czechia
| | - Filip Frühauf
- Department of Obstetrics and Gynecology, First Faculty of Medicine, Charles University and General University Hospital in Prague, Prague, Czechia
| | - Roman Kocian
- Department of Obstetrics and Gynecology, First Faculty of Medicine, Charles University and General University Hospital in Prague, Prague, Czechia
| | - Tereza Indrielle-Kelly
- Department of Obstetrics and Gynecology, Burton Hospitals National Health System (NHS), West Midlands, United Kingdom
| | - David Cibula
- Department of Obstetrics and Gynecology, First Faculty of Medicine, Charles University and General University Hospital in Prague, Prague, Czechia
| |
Collapse
|
5
|
Arezzo F, Cormio G, La Forgia D, Santarsiero CM, Mongelli M, Lombardi C, Cazzato G, Cicinelli E, Loizzi V. A machine learning approach applied to gynecological ultrasound to predict progression-free survival in ovarian cancer patients. Arch Gynecol Obstet 2022; 306:2143-2154. [PMID: 35532797 PMCID: PMC9633520 DOI: 10.1007/s00404-022-06578-1] [Citation(s) in RCA: 9] [Impact Index Per Article: 4.5] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/21/2022] [Accepted: 04/12/2022] [Indexed: 02/05/2023]
Abstract
In a growing number of social and clinical scenarios, machine learning (ML) is emerging as a promising tool for implementing complex multi-parametric decision-making algorithms. Regarding ovarian cancer (OC), despite the standardization of features that can support the discrimination of ovarian masses into benign and malignant, there is a lack of accurate predictive modeling based on ultrasound (US) examination for progression-free survival (PFS). This retrospective observational study analyzed patients with epithelial ovarian cancer (EOC) who were followed in a tertiary center from 2018 to 2019. Demographic features, clinical characteristics, information about the surgery and post-surgery histopathology were collected. Additionally, we recorded data about US examinations according to the International Ovarian Tumor Analysis (IOTA) classification. Our study aimed to realize a tool to predict 12 month PFS in patients with OC based on a ML algorithm applied to gynecological ultrasound assessment. Proper feature selection was used to determine an attribute core set. Three different machine learning algorithms, namely Logistic Regression (LR), Random Forest (RFF), and K-nearest neighbors (KNN), were then trained and validated with five-fold cross-validation to predict 12 month PFS. Our analysis included n. 64 patients and 12 month PFS was achieved by 46/64 patients (71.9%). The attribute core set used to train machine learning algorithms included age, menopause, CA-125 value, histotype, FIGO stage and US characteristics, such as major lesion diameter, side, echogenicity, color score, major solid component diameter, presence of carcinosis. RFF showed the best performance (accuracy 93.7%, precision 90%, recall 90%, area under receiver operating characteristic curve (AUROC) 0.92). We developed an accurate ML model to predict 12 month PFS.
Collapse
Affiliation(s)
- Francesca Arezzo
- Department of Biomedical Sciences and Human Oncology, Obstetrics and Gynecology Unit, University of Bari “Aldo Moro”, Piazza Giulio Cesare 11, 70124 Bari, Italy
| | - Gennaro Cormio
- Department of Biomedical Sciences and Human Oncology, Obstetrics and Gynecology Unit, University of Bari “Aldo Moro”, Piazza Giulio Cesare 11, 70124 Bari, Italy
| | - Daniele La Forgia
- Department of Breast Radiology, Giovanni Paolo II I.R.C.C.S. Cancer Institute, via Orazio Flacco 65, 70124 Bari, Italy
| | - Carla Mariaflavia Santarsiero
- Department of Biomedical Sciences and Human Oncology, Obstetrics and Gynecology Unit, University of Bari “Aldo Moro”, Piazza Giulio Cesare 11, 70124 Bari, Italy
| | - Michele Mongelli
- Department of Biomedical Sciences and Human Oncology, Obstetrics and Gynecology Unit, University of Bari “Aldo Moro”, Piazza Giulio Cesare 11, 70124 Bari, Italy
| | - Claudio Lombardi
- Department of Biomedical Sciences and Human Oncology, Obstetrics and Gynecology Unit, University of Bari “Aldo Moro”, Piazza Giulio Cesare 11, 70124 Bari, Italy
| | - Gerardo Cazzato
- Department of Emergency and Organ Transplantation, Pathology Section, University of Bari “Aldo Moro”, Piazza Giulio Cesare 11, 70124 Bari, Italy
| | - Ettore Cicinelli
- Department of Biomedical Sciences and Human Oncology, Obstetrics and Gynecology Unit, University of Bari “Aldo Moro”, Piazza Giulio Cesare 11, 70124 Bari, Italy
| | - Vera Loizzi
- Interdisciplinar Department of Medicine, Obstetrics and Gynecology Unit, University of Bari “Aldo Moro”, Piazza Giulio Cesare 11, 70124 Bari, Italy
| |
Collapse
|